CN111552865A - User interest portrait method and related equipment - Google Patents

Abstract

The present invention provides a user interest portrait method and related equipment. The user interest portrait method judges whether multiple websites have user registration information according to the identification information, and obtains multiple target websites; generates user registration feature vectors according to the judgment results of whether multiple websites have user registration information; adopts a clustering method Determine the first probability value of each interest tag of the user according to the user's registration feature vector; crawl multiple target named entities of the user from each target website; use the trained neural network to calculate the second probability value of each interest tag ; Calculate the third probability value of each interest tag based on a statistical method; determine the maximum value among the first probability value, the second probability value and the third probability value of each interest tag as the target probability value of the interest tag; An interest tag whose target probability value is greater than the first preset threshold is determined as an interest tag of the user. The present invention improves the accuracy of extracting interest tags of users.

Description

User interest portrait method and related equipment

technical field

The present invention relates to the technical field of entity recognition, and in particular, to a user interest portrait method, device, computer equipment and computer-readable storage medium.

Background technique

Interests and hobbies in user interest portraits are important data in modern financial scenarios, and are widely used in many fields such as marketing, services, and even risk control.

User interest portraits need to extract users’ interest tags (such as travel, programming learning, etc.). The existing user interest portrait methods extract users’ interest tags based on the social and usage data of users on a certain platform. Defects lead to low accuracy in extracting user interest tags. How to accurately extract users' interest tags has become an urgent problem to be solved.

SUMMARY OF THE INVENTION

In view of the above content, it is necessary to propose a user interest portrait method, apparatus, computer equipment and computer-readable storage medium, which can extract the user's interest tag according to the user's registration information on various websites.

A first aspect of the present application provides a user interest portrait method, and the user interest portrait method includes:

Obtain identifying information from multiple websites, multiple interest tags, and users;

Determine whether the user's registration information exists in the plurality of websites according to the identification information, and obtain a plurality of target websites in which the user's registration information exists;

generating the registration feature vector of the user according to the judgment result of whether the registration information of the user exists in the plurality of websites;

Using a clustering method to determine the first probability value of each interest tag of the user according to the registration feature vector of the user;

Crawl a plurality of target named entities for said user from each target website;

Using the trained neural network to calculate the second probability value of each interest tag according to the multiple target named entities and the target website to which each target named entity belongs;

Calculate the third probability value of each interest tag based on a statistical method;

determining the maximum value among the first probability value, the second probability value and the third probability value of each interest tag as the target probability value of the interest tag;

An interest tag whose target probability value is greater than the first preset threshold is determined as the user's interest tag.

In another possible implementation manner, the determining, according to the identification information, whether the multiple websites have the registration information of the user includes:

searching for the identifying information at a designated website among the plurality of websites;

If the identification information is included in the search result of the designated website, the designated website has the registration information of the user;

If the search result of the designated website does not include the identification information, the designated website does not have the registration information of the user.

In another possible implementation manner, the determining, according to the identification information, whether the multiple websites have the registration information of the user includes:

Query the registration information of the user from an interface authorized by a designated website among the plurality of websites according to the identification information;

If the designated website returns the user's registration information, the designated website has the user's registration information;

If the designated website does not return the user's registration information or the return value is empty, the designated website does not have the user's registration information.

In another possible implementation manner, the adopting a clustering method to determine the first probability value of each interest tag of the user according to the registration feature vector of the user includes:

Get multiple first historical users;

Clustering the plurality of first historical users according to the registration feature vectors of the plurality of first historical users to obtain a plurality of user clusters and the center vector of each user cluster;

Determine the target user cluster to which the user belongs according to the distance between the registration feature vector of the user and the center vector of each user cluster;

Determine the mean value of the probability values of each target user in the target user cluster about the specified interest label as the first probability value of the user's specified interest label, or determine the probability value of the specified interest label in the target user cluster greater than The ratio of the number of target users at the second preset threshold to the total number of target users in the target user cluster is determined as the first probability value of the specified interest tag of the user.

In another possible implementation manner, calculating the second probability value of each interest tag according to the multiple target named entities and the target website to which each target named entity belongs by using the trained neural network includes:

Encoding each target named entity and the target website to which the target named entity belongs as a feature vector of the target named entity;

Input the feature vector of each target named entity into the trained neural network to obtain the probability value of each interest label corresponding to the target named entity;

The mean value of the probability values of each interest tag corresponding to the multiple named entities is calculated to obtain the second probability value of the interest tag.

In another possible implementation manner, the calculating the third probability value of each interest tag based on a statistical method includes:

Acquiring multiple second historical users with registration information on the multiple target websites, and the user interest portrait of each second historical user includes multiple tags of the second historical user;

Counting the first number of second historical users with the interest tag in the user interest portrait;

Counting the second number of the plurality of second historical users;

A ratio of the first number to the second number is calculated, and the ratio of the first number to the second number is used as the third probability value.

In another possible implementation manner, before judging whether the multiple websites have the registration information of the user according to the identification information, the user interest portrait method further includes: obtaining the authorization of the user.

A second aspect of the present application provides a user interest portrait device, and the user interest portrait device includes:

The acquisition module is used to acquire the identification information of multiple websites, multiple interest tags and users;

a judging module, configured to judge whether the multiple websites have the registration information of the user according to the identification information, and obtain multiple target websites that have the registration information of the user;

a generating module, configured to generate the registration feature vector of the user according to the judgment result of whether the registration information of the user exists in the multiple websites;

a first determination module, configured to adopt a clustering method to determine the first probability value of each interest tag of the user according to the registration feature vector of the user;

a crawling module for crawling multiple target named entities of the user from each target website;

The first calculation module is used to calculate the second probability value of each interest tag according to the plurality of target named entities and the target website to which each target named entity belongs with the trained neural network;

The second calculation module is used to calculate the third probability value of each interest tag based on a statistical method;

a second determination module, configured to determine the maximum value among the first probability value, the second probability value and the third probability value of each interest tag as the target probability value of the interest tag;

The third determination module is configured to determine an interest tag whose target probability value is greater than the first preset threshold as an interest tag of the user.

A third aspect of the present application provides a computer device, the computer device includes a processor, and the processor is configured to implement the user interest portrait method when executing a computer program stored in a memory.

A fourth aspect of the present application provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, implements the user interest portrait method.

The present invention determines the interest label of the user through the website associated with the user's interest and the target named entity of the user in the target website, which can improve the accuracy of identifying the user's interest label; The first probability value of the interest label, the second probability value of the interest label obtained by the neural network, and the third probability value of the interest label obtained by statistics can determine the target probability value of the interest label, which can reduce the risk of deviation. Therefore, the present invention realizes that the user's interest tags are extracted according to the user's registration information on various websites, and the accuracy of extracting the user's interest tags is improved.

Description of drawings

FIG. 1 is a flowchart of a user interest portrait method provided by an embodiment of the present invention.

FIG. 2 is a structural diagram of a user interest portrait device provided by an embodiment of the present invention.

FIG. 3 is a schematic diagram of a computer device provided by an embodiment of the present invention.

Detailed ways

In order to more clearly understand the above objects, features and advantages of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments may be combined with each other in the case of no conflict.

In the following description, many specific details are set forth in order to facilitate a full understanding of the present invention, and the described embodiments are only some, but not all, embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. The terms used herein in the description of the present invention are for the purpose of describing specific embodiments only, and are not intended to limit the present invention.

Preferably, the user interest profile method of the present invention is applied in one or more computer devices. The computer device is a device that can automatically perform numerical calculation and/or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to a microprocessor, an application specific integrated circuit (ASIC), Programmable Gate Array (Field-Programmable Gate Array, FPGA), Digital Signal Processor (Digital Signal Processor, DSP), embedded devices, etc.

The computer equipment may be a desktop computer, a notebook computer, a palmtop computer, a cloud server and other computing equipment. The computer device can perform human-computer interaction with the user through a keyboard, a mouse, a remote control, a touch pad or a voice control device.

Example 1

FIG. 1 is a flowchart of a user interest portrait method provided by Embodiment 1 of the present invention. The user interest portrait method is applied to computer equipment, and is used for extracting user interest tags according to the user's registration information on various websites.

As shown in Figure 1, the user interest portrait method includes:

101. Acquire identification information of multiple websites, multiple interest tags, and users.

In a specific embodiment, the plurality of websites may include NetEase Cloud Music, NetEase Cloud Classroom, Baidu Tieba, CSDN, Weibo, Xiaohongshu, Ctrip, and the like.

The plurality of interest tags may include fitness, education, mother and child, travel, and the like. Users with different interests may be using websites associated with their interests (that is, the websites associated with the interests of the users have user registration information), and there is an association between the interests of the users and the websites registered by the users. For example, travel is associated with Ctrip, and education is associated with NetEase Cloud Classroom.

The identification information input by the user or the identification information of the user transmitted by the user identification device may be received.

In a specific embodiment, the identification information includes a mobile phone number, an ID card number, an encrypted mobile phone number or an encrypted ID card number.

For example, the mobile phone number or ID number input by the user through the keyboard, or the ID number of the user transmitted by the text recognition device can be received, and the text recognition device can recognize the ID number on the ID card of the user. The mobile phone number can be encrypted by hash encryption or MD5 encryption algorithm to obtain the encrypted mobile phone number, and the ID card number can be encrypted to obtain the encrypted ID card number.

In another embodiment, the identification information may further include fingerprint information, iris information, or face information, and the like.

102. Determine, according to the identification information, whether the multiple websites have the user's registration information, and obtain multiple target websites that have the user's registration information.

In a specific embodiment, the determining whether there is the user's registration information on the multiple websites according to the identification information includes:

Query the registration information of the user from an interface authorized by a designated website among the plurality of websites according to the identification information;

If the designated website returns the user's registration information, the designated website has the user's registration information;

If the designated website does not return the user's registration information or the return value is empty, the designated website does not have the user's registration information.

For example, query the registration information of user A from the registration information query interface of CSDN (the query parameter is the phone number of user A); if CSDN returns the registration information of user A (such as user A's registration time, registration status, user name, etc.), Then the CSDN has the registration information of user A.

In another embodiment, the determining, according to the identification information, whether the multiple websites have registration information of the user includes:

registering a new account with a designated one of the plurality of websites using the identifying information;

If the designated website prompts that the user has registered, the designated website has the registration information of the user;

If the designated website prompts to input registration verification information, the designated website does not have the user's registration information.

For example, user A's phone number can be used to request CSDN to register a new account; if CSDN prompts to enter registration verification information (such as the verification code of the phone number issued by CSDN to user A), then CSDN does not have user A's registration information.

In another embodiment, the determining, according to the identification information, whether the multiple websites have registration information of the user includes:

searching for the identifying information at a designated website among the plurality of websites;

If the identification information is included in the search result of the designated website, the designated website has the registration information of the user;

If the search result of the designated website does not include the identification information, the designated website does not have the registration information of the user.

103. Generate a registration feature vector of the user according to a result of determining whether the user's registration information exists on the multiple websites.

For example, the generated registration feature vector of user A is (1, 1, 0, 1, 0), where 1 in the first dimension from left to right indicates that NetEase Cloud Music has user A's registration information; the second The 1 of the dimension means that the registration information of user A exists in Baidu Tieba; the 0 of the third dimension means that the registration information of user A does not exist in CSDN; the 1 of the fourth dimension means that the registration information of user A exists in Weibo; the 0 means that the registration information of user A does not exist in Xiaohongshu.

104. Use a clustering method to determine a first probability value of each interest tag of the user according to the registered feature vector of the user.

In a specific embodiment, the adopting a clustering method to determine the first probability value of each interest tag of the user according to the registration feature vector of the user includes:

(1) Acquire multiple first historical users.

(2) Clustering the plurality of first historical users according to the registration feature vectors of the plurality of first historical users to obtain a plurality of user clusters and a center vector of each user cluster.

(3) Determine the target user cluster to which the user belongs according to the distance between the registered feature vector of the user and the center vector of each user cluster. For example, two user clusters (respectively the first user cluster and the second user cluster) are obtained by clustering, the Euclidean distance between the user's registration feature vector and the center vector of the first user cluster is num1, and the user's registration feature vector and the first user cluster are num1. The Euclidean distance of the center vector of the two user clusters is num2, and if num1 is greater than num2, the second user cluster is determined as the target user cluster.

(4) Determine the mean value of the probability values of each target user in the target user cluster about the specified interest label as the first probability value of the user's specified interest label, or determine the average value of the specified interest label in the target user cluster as the first probability value of the specified interest label. The ratio of the number of target users whose probability value is greater than the second preset threshold to the total number of target users in the target user cluster is determined as the first probability value of the specified interest tag of the user. For example, if the target user cluster includes 3 users, the specified interest label is travel, and the probability values of the travel interest labels of the three users are 0.5, 0.6, and 0.4, respectively, then the first probability value of the user's travel interest label is 0.5. The second preset threshold is a preset value adjusted according to experimental data.

105. Crawl multiple target named entities of the user from each target website.

A plurality of webpage texts of the user can be crawled from each target website, and the plurality of webpage texts of the target website include the social information and behavior information of the user on the target website, extracted from the plurality of webpage texts of the target website. Multiple target named entities for this target website. For example, the target website is NetEase Cloud Music, and the multiple webpage texts include the playlists that User A follows or shares. Named entity extraction is performed from multiple webpage texts, and the obtained multiple target named entities are folk ballads, campuses, etc. (folk ballads are extracted from the , campus users were found to be generally fond of "travel"). Another example is that the target website is Xiaohongshu, and multiple webpage texts include the overseas shopping experience that User B follows or shares. Named entities are extracted from multiple webpage texts, and the obtained multiple target named entities are milk powder, baby carriage, etc. Users who extracted milk powder and strollers were found to be generally inclined to "mother and baby"). For another example, the target website is NetEase Cloud Classroom, and multiple webpage texts include video introductions that User B follows or shares. Named entity extraction is performed from multiple webpage texts, and the obtained multiple target named entities are JAVA, SPRING, etc. JAVA, SPRING users are found to generally tend to "programming education").

106. Use the trained neural network to calculate the second probability value of each interest tag according to the multiple target named entities and the target website to which each target named entity belongs.

In a specific embodiment, calculating the second probability value of each interest tag according to the multiple target named entities and the target website to which each target named entity belongs by using the trained neural network includes:

Encoding each target named entity and the target website to which the target named entity belongs as a feature vector of the target named entity;

Input the feature vector of each target named entity into the trained neural network to obtain the probability value of each interest label corresponding to the target named entity;

The mean value of the probability values of each interest tag corresponding to the multiple named entities is calculated to obtain the second probability value of the interest tag.

For example, the two named entities are JAVA and SPRING. The probability value of JAVA corresponding to "programming education" (interest label) is 0.9, and the probability value of SPRING corresponding to programming education (interest label) is 0.7, then "programming education" (interest label) has a probability value of 0.9. ) has a second probability value of 0.8.

The feature vector encoding each target named entity and the target website to which the target named entity belongs as the target named entity includes:

Encode the target named entity into a first intermediate vector according to a preset encoder (such as one-hot encoder, word2vec encoder);

According to the preset encoder, the target website to which the target named entity belongs is encoded as a second intermediate vector;

Connect the first intermediate vector and the second intermediate vector, or multiply the elements of the first intermediate vector and the second intermediate vector to obtain the feature vector of the target named entity.

Training the neural network may include:

Obtain a training sample and a label of the training sample, where the training sample includes a target named entity and the target website code to which the target named entity belongs;

The target named entity is encoded into a first vector according to a preset encoding table, and the target website to which the target named entity belongs is encoded into a second vector according to the preset encoding table;

Splicing the first vector and the second vector to obtain the feature vector of the target named entity;

Input the feature vector of the target named entity into the neural network of the initialization parameter value to obtain the output vector;

According to the output vector and the label of the training sample, the parameter values of the neural network are optimized through a back-propagation algorithm.

107. Calculate a third probability value of each interest tag based on a statistical method.

The calculation of the third probability value of each interest tag based on the statistical method includes:

Acquiring multiple second historical users with registration information on the multiple target websites, and the user interest portrait of each second historical user includes multiple tags of the second historical user;

Counting the first number of second historical users with the interest tag in the user interest portrait;

Counting the second number of the plurality of second historical users;

A ratio of the first number to the second number is calculated, and the ratio of the first number to the second number is used as the third probability value.

For example, obtain 4 second historical users (respectively, user A, user B, user C, and user D) with registration information in NetEase Cloud Music and Ctrip; count the first users with "travel" (interest tag) in the user's interest portrait. The first number of second historical users is 3; the second number of second historical users is 4; the third probability value of "travel" (interest tag) is 0.75.

108. Determine the maximum value among the first probability value, the second probability value and the third probability value of each interest tag as the target probability value of the interest tag.

For example, the first probability value of "Travel" (interest tag) is 0.65, the third probability value of "Travel" (interest tag) is 0.70, and the third probability value of "Travel" (interest tag) is 0.75, then 0.75 The target probability value determined as "tour" (interest label).

109. Determine an interest tag with a target probability value greater than a first preset threshold as an interest tag of the user.

For example, the target probability value of "travel" (interest label) is 0.75, the target probability value of "programming education" (interest label) is 0.85, and the first preset threshold is 0.80, then "programming education" is determined as the the user's interest tags.

The user interest profile method of Embodiment 1 determines the user's interest tag by using a website associated with the user's interest and the user's target named entity in the target website, which can improve the accuracy of identifying the user's interest tag; The first probability value of the interest label obtained by the clustering method, the second probability value of the interest label obtained by the neural network, and the third probability value of the interest label obtained by statistics can determine the target probability value of the interest label, which can reduce the occurrence of Risk of Bias. The first embodiment extracts the user's interest tags according to the user's registration information on various websites, which improves the accuracy of extracting the user's interest tags, and uses the extracted user's interest tags to describe the user's interest portrait, which improves the accuracy of describing the user's interest portrait. .

In another embodiment, before judging whether the multiple websites have registration information of the user according to the identification information, the user interest portrait method further includes: obtaining authorization from the user.

Before determining whether the user's registration information exists in the plurality of websites according to the identification information, an authorization option box may be issued to the user, and an authorization option checked by the user in the authorization option box may be received.

Embodiment 2

FIG. 2 is a structural diagram of a user interest portrait device according to Embodiment 2 of the present invention. The user interest portrait device 20 is applied to computer equipment. The user interest portrait device 20 is configured to extract the user's interest tags according to the user's registration information on various websites.

As shown in FIG. 2, the user interest portrait device 20 may include an acquisition module 201, a judgment module 202, a generation module 203, a first determination module 204, a crawling module 205, a first calculation module 206, a second calculation module 207, The second determination module 208 and the third determination module 209 .

The acquiring module 201 is used for acquiring multiple websites, multiple interest tags and user identification information.

In a specific embodiment, the plurality of websites may include NetEase Cloud Music, NetEase Cloud Classroom, Baidu Tieba, CSDN, Weibo, Xiaohongshu, Ctrip, and the like.

The plurality of interest tags may include fitness, education, mother and child, travel, and the like. Users with different interests may be using websites associated with their interests (that is, the websites associated with the interests of the users have user registration information), and there is an association between the interests of the users and the websites registered by the users. For example, travel is associated with Ctrip, and education is associated with NetEase Cloud Classroom.

The identification information input by the user or the identification information of the user transmitted by the user identification device may be received.

In a specific embodiment, the identification information includes a mobile phone number, an ID card number, an encrypted mobile phone number or an encrypted ID card number.

For example, the mobile phone number or ID number input by the user through the keyboard, or the ID number of the user transmitted by the text recognition device can be received, and the text recognition device can recognize the ID number on the ID card of the user. The mobile phone number can be encrypted by hash encryption or MD5 encryption algorithm to obtain the encrypted mobile phone number, and the ID card number can be encrypted to obtain the encrypted ID card number.

In another embodiment, the identification information may further include fingerprint information, iris information, or face information, and the like.

The judgment module 202 is configured to judge whether the user's registration information exists in the plurality of websites according to the identification information, and obtain a plurality of target websites in which the user's registration information exists.

In a specific embodiment, the determining whether there is the user's registration information on the multiple websites according to the identification information includes:

Query the registration information of the user from an interface authorized by a designated website among the plurality of websites according to the identification information;

If the designated website returns the user's registration information, the designated website has the user's registration information;

If the designated website does not return the user's registration information or the return value is empty, the designated website does not have the user's registration information.

For example, query the registration information of user A from the registration information query interface of CSDN (the query parameter is the phone number of user A); if CSDN returns the registration information of user A (such as user A's registration time, registration status, user name, etc.), Then the registration information of user A exists in CSDN.

In another embodiment, the determining, according to the identification information, whether the multiple websites have registration information of the user includes:

registering a new account with a designated one of the plurality of websites using the identifying information;

If the designated website prompts that the user has registered, the designated website has the registration information of the user;

If the designated website prompts to input registration verification information, the designated website does not have the user's registration information.

For example, user A's phone number can be used to request CSDN to register a new account; if CSDN prompts to enter registration verification information (such as the verification code of the phone number issued by CSDN to user A), then CSDN does not have user A's registration information.

In another embodiment, the determining, according to the identification information, whether the multiple websites have registration information of the user includes:

searching for the identifying information at a designated website among the plurality of websites;

If the identification information is included in the search result of the designated website, the designated website has the registration information of the user;

If the search result of the designated website does not include the identification information, the designated website does not have the registration information of the user.

The generating module 203 is configured to generate the registration feature vector of the user according to the judgment result of whether the registration information of the user exists in the multiple websites.

For example, the generated registration feature vector of user A is (1, 1, 0, 1, 0), where 1 in the first dimension from left to right indicates that NetEase Cloud Music has user A's registration information; the second The 1 of the dimension means that the registration information of user A exists in Baidu Tieba; the 0 of the third dimension means that the registration information of user A does not exist in CSDN; the 1 of the fourth dimension means that the registration information of user A exists in Weibo; the 0 means that the registration information of user A does not exist in Xiaohongshu.

The first determining module 204 is configured to adopt a clustering method to determine the first probability value of each interest tag of the user according to the registration feature vector of the user.

In a specific embodiment, the adopting a clustering method to determine the first probability value of each interest tag of the user according to the registration feature vector of the user includes:

(1) Acquire multiple first historical users.

(2) Clustering the plurality of first historical users according to the registration feature vectors of the plurality of first historical users to obtain a plurality of user clusters and a center vector of each user cluster.

(3) Determine the target user cluster to which the user belongs according to the distance between the registered feature vector of the user and the center vector of each user cluster. For example, two user clusters (respectively the first user cluster and the second user cluster) are obtained by clustering, the Euclidean distance between the user's registration feature vector and the center vector of the first user cluster is num1, and the user's registration feature vector and the first user cluster are num1. The Euclidean distance of the center vector of the two user clusters is num2, and if num1 is greater than num2, the second user cluster is determined as the target user cluster.

(4) Determine the mean value of the probability values of each target user in the target user cluster about the specified interest label as the first probability value of the user's specified interest label, or determine the average value of the specified interest label in the target user cluster as the first probability value of the specified interest label. The ratio of the number of target users whose probability value is greater than the second preset threshold to the total number of target users in the target user cluster is determined as the first probability value of the specified interest tag of the user. For example, if the target user cluster includes 3 users, the specified interest label is travel, and the probability values of the travel interest labels of the three users are 0.5, 0.6, and 0.4, respectively, then the first probability value of the user's travel interest label is 0.5. The second preset threshold is a preset value adjusted according to experimental data.

The crawling module 205 is configured to crawl a plurality of target named entities of the user from each target website.

A plurality of webpage texts of the user can be crawled from each target website, and the plurality of webpage texts of the target website include the social information and behavior information of the user on the target website, extracted from the plurality of webpage texts of the target website. Multiple target named entities for this target website. For example, the target website is NetEase Cloud Music, and the multiple webpage texts include the playlists that User A follows or shares. Named entity extraction is performed from multiple webpage texts, and the obtained multiple target named entities are folk ballads, campuses, etc. (folk ballads are extracted from the , campus users were found to be generally fond of "travel"). Another example is that the target website is Xiaohongshu, and multiple webpage texts include the overseas shopping experience that User B follows or shares. Named entities are extracted from multiple webpage texts, and the obtained multiple target named entities are milk powder, baby carriage, etc. Users who extracted milk powder and strollers were found to be generally inclined to "mother and baby"). For another example, the target website is NetEase Cloud Classroom, and multiple webpage texts include video introductions that User B follows or shares. Named entity extraction is performed from multiple webpage texts, and the obtained multiple target named entities are JAVA, SPRING, etc. JAVA, SPRING users are found to generally tend to "programming education").

The first calculation module 206 is configured to use the trained neural network to calculate the second probability value of each interest tag according to the plurality of target named entities and the target website to which each target named entity belongs.

In a specific embodiment, calculating the second probability value of each interest tag according to the multiple target named entities and the target website to which each target named entity belongs by using the trained neural network includes:

Encoding each target named entity and the target website to which the target named entity belongs as a feature vector of the target named entity;

Input the feature vector of each target named entity into the trained neural network to obtain the probability value of each interest label corresponding to the target named entity;

The mean value of the probability values of each interest tag corresponding to the multiple named entities is calculated to obtain the second probability value of the interest tag. For example, the two named entities are JAVA and SPRING. The probability value of JAVA corresponding to "programming education" (interest label) is 0.9, and the probability value of SPRING corresponding to programming education (interest label) is 0.7, then "programming education" (interest label) has a probability value of 0.9. ) has a second probability value of 0.8.

The feature vector encoding each target named entity and the target website to which the target named entity belongs as the target named entity includes:

Encode the target named entity into a first intermediate vector according to a preset encoder (such as one-hot encoder, word2vec encoder);

According to the preset encoder, the target website to which the target named entity belongs is encoded as the second intermediate vector;

Connect the first intermediate vector and the second intermediate vector, or multiply the elements of the first intermediate vector and the second intermediate vector to obtain the feature vector of the target named entity.

Training the neural network may include:

Obtain a training sample and a label of the training sample, where the training sample includes a target named entity and the target website code to which the target named entity belongs;

The target named entity is encoded into a first vector according to a preset encoding table, and the target website to which the target named entity belongs is encoded into a second vector according to the preset encoding table;

Splicing the first vector and the second vector to obtain the feature vector of the target named entity;

Input the feature vector of the target named entity into the neural network of the initialization parameter value to obtain the output vector;

According to the output vector and the label of the training sample, the parameter values of the neural network are optimized through a back-propagation algorithm.

The second calculation module 207 is configured to calculate the third probability value of each interest tag based on a statistical method.

The calculation of the third probability value of each interest tag based on the statistical method includes:

Acquiring multiple second historical users with registration information on the multiple target websites, and the user interest portrait of each second historical user includes multiple tags of the second historical user;

Counting the first number of second historical users with the interest tag in the user interest portrait;

Counting the second number of the plurality of second historical users;

A ratio of the first number to the second number is calculated, and the ratio of the first number to the second number is used as the third probability value.

For example, obtain 4 second historical users (respectively, user A, user B, user C, and user D) with registration information in NetEase Cloud Music and Ctrip; count the first users with "travel" (interest tag) in the user's interest portrait. The first number of second historical users is 3; the second number of second historical users is 4; the third probability value of "travel" (interest tag) is 0.75.

The second determination module 208 is configured to determine the maximum value among the first probability value, the second probability value and the third probability value of each interest tag as the target probability value of the interest tag.

For example, the first probability value of "Travel" (interest tag) is 0.65, the third probability value of "Travel" (interest tag) is 0.70, and the third probability value of "Travel" (interest tag) is 0.75, then 0.75 The target probability value determined as "tour" (interest label).

The third determination module 209 is configured to determine an interest tag whose target probability value is greater than the first preset threshold as an interest tag of the user.

For example, the target probability value of "travel" (interest label) is 0.75, the target probability value of "programming education" (interest label) is 0.85, and the first preset threshold is 0.80, then "programming education" is determined as the the user's interest tags.

The user interest portrait device 20 of the second embodiment determines the interest label of the user through the website associated with the user's interest and the target named entity of the user in the target website, which can improve the accuracy rate of identifying the user's interest label ; The first probability value of the interest label obtained by the clustering method, the second probability value of the interest label obtained by the neural network and the third probability value of the interest label obtained based on statistics can determine the target probability value of the interest label, which can reduce the Risk of Bias. The second embodiment extracts the user's interest tags according to the user's registration information on various websites, improves the accuracy of extracting the user's interest tags, and uses the extracted user's interest tags to describe the user's interest portrait, which improves the accuracy of describing the user's interest portrait. .

In another embodiment, the acquiring module is further configured to acquire the authorization of the user before determining whether the multiple websites have the registration information of the user according to the identification information.

Before determining whether the user's registration information exists in the plurality of websites according to the identification information, an authorization option box may be issued to the user, and an authorization option checked by the user in the authorization option box may be received.

Embodiment 3

This embodiment provides a computer-readable storage medium, where a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, implements the steps in the above embodiment of the user interest portrait method, for example, as shown in FIG. 1 . Steps 101-109:

101. Obtain identification information of multiple websites, multiple interest tags and users;

102. Determine whether the multiple websites have the registration information of the user according to the identification information, and obtain multiple target websites that have the registration information of the user;

103. Generate a registration feature vector of the user according to the judgment result of whether the registration information of the user exists in the multiple websites;

104, adopting a clustering method to determine the first probability value of each interest tag of the user according to the registration feature vector of the user;

105. Crawl multiple target named entities of the user from each target website;

106, calculating the second probability value of each interest tag with the trained neural network according to the multiple target named entities and the target website to which each target named entity belongs;

107. Calculate the third probability value of each interest tag based on a statistical method;

108, determining the maximum value among the first probability value, the second probability value and the third probability value of each interest tag as the target probability value of the interest tag;

109. Determine an interest tag with a target probability value greater than a first preset threshold as an interest tag of the user.

Alternatively, when the computer program is executed by the processor, the functions of the modules in the above-mentioned apparatus embodiments are implemented, for example, the modules 201-209 in FIG. 2:

an acquisition module 201, configured to acquire multiple websites, multiple interest tags and user identification information;

A judgment module 202, configured to judge whether the multiple websites have the registration information of the user according to the identification information, and obtain multiple target websites that have the registration information of the user;

The generating module 203 is used for generating the registration feature vector of the user according to the judgment result of whether the registration information of the user exists in the multiple websites;

a first determining module 204, configured to adopt a clustering method to determine the first probability value of each interest tag of the user according to the registration feature vector of the user;

A crawling module 205, configured to crawl a plurality of target named entities of the user from each target website;

The first calculation module 206 is used to calculate the second probability value of each interest tag according to the plurality of target named entities and the target website to which each target named entity belongs with the trained neural network;

The second calculation module 207 is configured to calculate the third probability value of each interest tag based on a statistical method;

The second determination module 208 is configured to determine the maximum value among the first probability value, the second probability value and the third probability value of each interest tag as the target probability value of the interest tag;

The third determination module 209 is configured to determine an interest tag whose target probability value is greater than the first preset threshold as an interest tag of the user.

Embodiment 4

FIG. 3 is a schematic diagram of a computer device according to Embodiment 3 of the present invention. The computer device 30 includes a memory 301, a processor 302, and a computer program 303 stored in the memory 301 and executable on the processor 302, such as a user interest profile program. When the processor 302 executes the computer program 303, the steps in the above embodiments of the user interest portrait method are implemented, for example, 101-109 shown in FIG. 1:

101. Obtain identification information of multiple websites, multiple interest tags and users;

102. Determine whether the multiple websites have the registration information of the user according to the identification information, and obtain multiple target websites that have the registration information of the user;

103. Generate a registration feature vector of the user according to the judgment result of whether the registration information of the user exists in the multiple websites;

104, adopting a clustering method to determine the first probability value of each interest tag of the user according to the registration feature vector of the user;

105. Crawl multiple target named entities of the user from each target website;

106, calculating the second probability value of each interest tag with the trained neural network according to the multiple target named entities and the target website to which each target named entity belongs;

107. Calculate the third probability value of each interest tag based on a statistical method;

108, determining the maximum value among the first probability value, the second probability value and the third probability value of each interest tag as the target probability value of the interest tag;

109. Determine an interest tag with a target probability value greater than a first preset threshold as an interest tag of the user.

Alternatively, when the computer program is executed by the processor, the functions of the modules in the above-mentioned apparatus embodiments are implemented, for example, the modules 201-209 in FIG. 2:

an acquisition module 201, configured to acquire multiple websites, multiple interest tags and user identification information;

A judgment module 202, configured to judge whether the multiple websites have the registration information of the user according to the identification information, and obtain multiple target websites that have the registration information of the user;

The generating module 203 is used for generating the registration feature vector of the user according to the judgment result of whether the registration information of the user exists in the multiple websites;

a first determining module 204, configured to adopt a clustering method to determine the first probability value of each interest tag of the user according to the registration feature vector of the user;

A crawling module 205, configured to crawl a plurality of target named entities of the user from each target website;

The first calculation module 206 is used to calculate the second probability value of each interest tag according to the plurality of target named entities and the target website to which each target named entity belongs with the trained neural network;

The second calculation module 207 is configured to calculate the third probability value of each interest tag based on a statistical method;

The second determination module 208 is configured to determine the maximum value among the first probability value, the second probability value and the third probability value of each interest tag as the target probability value of the interest tag;

The third determination module 209 is configured to determine an interest tag whose target probability value is greater than the first preset threshold as an interest tag of the user.

Exemplarily, the computer program 303 may be divided into one or more modules, and the one or more modules are stored in the memory 301 and executed by the processor 302 to complete the method. The one or more modules may be a series of computer program instruction segments capable of performing specific functions, and the instruction segments are used to describe the execution process of the computer program 303 in the computer device 30 . For example, the computer program 303 can be divided into an acquisition module 201, a judgment module 202, a generation module 203, a first determination module 204, a crawling module 205, a first calculation module 206, a second calculation module 207, The second determination module 208 and the third determination module 209, for the specific functions of each module, refer to the second embodiment.

Those skilled in the art can understand that the schematic diagram 3 is only an example of the computer device 30, and does not constitute a limitation on the computer device 30, and may include more or less components than the one shown, or combine some components, or different For example, the computer device 30 may also include input and output devices, network access devices, buses, and the like.

The so-called processor 302 may be a central processing unit (Central Processing Unit, CPU), and may also be other general-purpose processors, digital signal processors (Digital Signal Processors, DSP), application specific integrated circuits (Application Specific Integrated Circuit, ASIC), Field-Programmable Gate Array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor 302 can also be any conventional processor, etc. The processor 302 is the control center of the computer device 30, and uses various interfaces and lines to connect the entire computer device 30. various parts.

The memory 301 can be used to store the computer program 303, and the processor 302 implements the computer device by running or executing the computer program or module stored in the memory 301 and calling the data stored in the memory 301. 30 various functions. The memory 301 may mainly include a stored program area and a stored data area, wherein the stored program area may store an operating system, an application program required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the storage data area may Data and the like created according to the use of the computer device 30 are stored. In addition, the memory 301 may include non-volatile memory such as hard disk, internal memory, plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, Flash Card (Flash Card), at least A magnetic disk storage device, flash memory device, or other non-volatile solid state storage device.

If the modules integrated in the computer device 30 are implemented in the form of software functional modules and sold or used as independent products, they may be stored in a computer-readable storage medium. Based on this understanding, the present invention can implement all or part of the processes in the methods of the above embodiments, and can also be completed by instructing relevant hardware through a computer program, and the computer program can be stored in a computer-readable storage medium. When the program is executed by the processor, the steps of the foregoing method embodiments can be implemented. Wherein, the computer program includes computer program code, and the computer program code may be in the form of source code, object code, executable file or some intermediate form, and the like. The computer-readable medium may include: any entity or device capable of carrying the computer program code, a recording medium, a U disk, a removable hard disk, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM, Read-Only Memory) .

In the several embodiments provided by the present invention, it should be understood that the disclosed system, apparatus and method may be implemented in other manners. For example, the apparatus embodiments described above are only illustrative. For example, the division of the modules is only a logical function division, and there may be other division manners in actual implementation.

The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical modules, that is, may be located in one place, or may be distributed to multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution in this embodiment.

In addition, each functional module in each embodiment of the present invention may be integrated into one processing module, or each module may exist physically alone, or two or more modules may be integrated into one module. The above-mentioned integrated modules can be implemented in the form of hardware, or can be implemented in the form of hardware plus software function modules.

The above-mentioned integrated modules implemented in the form of software functional modules may be stored in a computer-readable storage medium. The above-mentioned software function modules are stored in a storage medium, and include several instructions to enable a computer device (which may be a personal computer, a server, or a network device, etc.) or a processor (processor) to execute the user interests described in the various embodiments of the present invention. Some steps of the portrait method.

It will be apparent to those skilled in the art that the present invention is not limited to the details of the above-described exemplary embodiments, but that the present invention may be embodied in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments are to be regarded in all respects as illustrative and not restrictive, and the scope of the invention is to be defined by the appended claims rather than the foregoing description, which are therefore intended to fall within the scope of the claims. All changes within the meaning and range of the equivalents of , are included in the present invention. Any reference signs in the claims shall not be construed as limiting the involved claim. Furthermore, it is clear that the word "comprising" does not exclude other modules or steps, and the singular does not exclude the plural. Several modules or means recited in the system claims can also be implemented by one module or means by means of software or hardware. The terms first, second, etc. are used to denote names and do not denote any particular order.

Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be Modifications or equivalent substitutions can be made without departing from the spirit and scope of the technical solutions of the present invention.

Claims (10)

1. a user interest portrait method, is characterized in that, described user interest portrait method comprises: Obtain identifying information from multiple websites, multiple interest tags, and users; Determine whether the user's registration information exists in the plurality of websites according to the identification information, and obtain a plurality of target websites in which the user's registration information exists; generating the registration feature vector of the user according to the judgment result of whether the registration information of the user exists in the plurality of websites; Using a clustering method to determine the first probability value of each interest tag of the user according to the registration feature vector of the user; Crawl a plurality of target named entities for said user from each target website; Using the trained neural network to calculate the second probability value of each interest tag according to the multiple target named entities and the target website to which each target named entity belongs; Calculate the third probability value of each interest tag based on a statistical method; determining the maximum value among the first probability value, the second probability value and the third probability value of each interest tag as the target probability value of the interest tag; An interest tag whose target probability value is greater than the first preset threshold is determined as the user's interest tag. 2. The user interest portrait method according to claim 1, wherein the determining whether the user's registration information exists in the plurality of websites according to the identification information comprises: searching for the identifying information at a designated website among the plurality of websites; If the identification information is included in the search result of the designated website, the designated website has the registration information of the user; If the search result of the designated website does not include the identification information, the designated website does not have the registration information of the user. 3. The user interest portrait method according to claim 1, wherein the determining whether the multiple websites have the registration information of the user according to the identification information comprises: Query the registration information of the user from an interface authorized by a designated website among the plurality of websites according to the identification information; If the designated website returns the user's registration information, the designated website has the user's registration information; If the designated website does not return the user's registration information or the return value is empty, the designated website does not have the user's registration information. 4. The user interest portrait method according to claim 1, wherein the adopting a clustering method to determine the first probability value of each interest tag of the user according to the registration feature vector of the user comprises: Get multiple first historical users; Clustering the plurality of first historical users according to the registration feature vectors of the plurality of first historical users to obtain a plurality of user clusters and the center vector of each user cluster; Determine the target user cluster to which the user belongs according to the distance between the registration feature vector of the user and the center vector of each user cluster; Determine the mean value of the probability values of each target user in the target user cluster about the specified interest label as the first probability value of the user's specified interest label, or determine the probability value of the specified interest label in the target user cluster greater than The ratio of the number of target users at the second preset threshold to the total number of target users in the target user cluster is determined as the first probability value of the specified interest tag of the user. 5. user interest portrait method as claimed in claim 1 is characterized in that, described with trained neural network according to described multiple target named entities and the target website to which each target named entity belongs to calculate the value of each interest label. The second probability value includes: Encoding each target named entity and the target website to which the target named entity belongs as a feature vector of the target named entity; Input the feature vector of each target named entity into the trained neural network to obtain the probability value of each interest label corresponding to the target named entity; The mean value of the probability values of each interest tag corresponding to the multiple named entities is calculated to obtain the second probability value of the interest tag. 6. The user interest portrait method according to claim 1, wherein the calculating the third probability value of each interest tag based on a statistical method comprises: Acquiring multiple second historical users with registration information on the multiple target websites, and the user interest portrait of each second historical user includes multiple tags of the second historical user; Counting the first number of second historical users with the interest tag in the user interest portrait; Counting the second number of the plurality of second historical users; A ratio of the first number to the second number is calculated, and the ratio of the first number to the second number is used as the third probability value. 7. The user interest portrait method according to any one of claims 1-6, characterized in that, before judging whether the user's registration information exists in the plurality of websites according to the identification information, the The user interest portrait method further includes: obtaining the authorization of the user. 8. A user interest portrait device, wherein the user interest portrait device comprises: The acquisition module is used to acquire the identification information of multiple websites, multiple interest tags and users; a judging module, configured to judge whether the multiple websites have the registration information of the user according to the identification information, and obtain multiple target websites that have the registration information of the user; a generating module, configured to generate the registration feature vector of the user according to the judgment result of whether the registration information of the user exists in the multiple websites; a first determination module, configured to adopt a clustering method to determine the first probability value of each interest tag of the user according to the registration feature vector of the user; a crawling module for crawling multiple target named entities of the user from each target website; The first calculation module is used to calculate the second probability value of each interest tag according to the plurality of target named entities and the target website to which each target named entity belongs with the trained neural network; The second calculation module is used to calculate the third probability value of each interest tag based on a statistical method; a second determination module, configured to determine the maximum value among the first probability value, the second probability value and the third probability value of each interest tag as the target probability value of the interest tag; The third determination module is configured to determine an interest tag whose target probability value is greater than the first preset threshold as an interest tag of the user. 9 . A computer device, characterized in that the computer device comprises a processor, and the processor is configured to execute a computer program stored in a memory to implement the user interest portrait method according to any one of claims 1 to 7 . 10. A computer-readable storage medium on which a computer program is stored, characterized in that, when the computer program is executed by a processor, the user according to any one of claims 1-7 is implemented Interest portrait method.

Images